OpenAI to cut off AI models for SpaceX-owned Cursor, escalating feud with Musk ↗
OpenAI plans to stop supplying its models directly to Cursor after the coding platform came under SpaceX ownership. OpenAI says it isn't confident SpaceX will use its technology within its terms - yep, the Musk-Altman feud has found another battlefield.
Anthropic quickly moved in the other direction, saying it would increase compute support for Claude inside Cursor. Cursor says OpenAI models represent only a small slice of its traffic, so the breakup might sting less than the headline suggests... or so it seems. (Reuters)
Sony Music, Warner sue Anthropic, alleging a "brazen campaign" of intellectual property theft ↗
Sony Music Publishing, Warner Chappell and other publishers have sued Anthropic and two of its co-founders, accusing the company of illegally torrenting, scraping and downloading copyrighted works used to train Claude.
The publishers say thousands of protected works were involved, with the broader complaint also targeting allegedly pirated books containing lyrics and sheet music. Anthropic says it disagrees with the claims and will fight them in court - another rather large log on AI's copyright bonfire. (TechCrunch)
Sharp rise in documented incidents of AI escaping users' control, research finds ↗
The Loss of Control Observatory recorded more than 300 documented incidents involving AI systems ignoring instructions, deceiving users or pursuing unwanted goals in a single month - nearly twice the previous month's total.
Reported examples include AI systems impersonating their human controllers and finding ways around rules requiring human approval. Not every incident caused serious damage, thankfully, but the trend is getting harder to wave away as agents gain more autonomy. (The Guardian)
China's Tencent releases new open-source AI model for coding, research tasks ↗
Tencent released a preview of Hy4, an open-source mixture-of-experts model aimed at software engineering, research and financial analysis. It packs 770 billion total parameters while activating about 49 billion for an individual text request.
Tencent plans to integrate it with products including CodeBuddy and WorkBuddy. The company also admits the preview can spend too long wrestling with complex questions and over-check its own answers - uncannily relatable for an AI model. (The Indian Express)
Nvidia's AI advantage is moving beyond the GPU ↗
Nvidia's competitive moat is increasingly about the entire AI system, not merely selling the GPU at its center. Its Vera Rubin architecture combines Rubin GPUs with the Vera CPU, inference acceleration plus specialized storage and networking hardware.
The idea is to move data around enormous AI systems more efficiently, where bottlenecks can waste serious amounts of compute. Nvidia says Vera has delivered up to a 3x improvement in some operations - the AI chip war is becoming a plumbing war, basically. (TechCrunch)
At TechBBQ, Europe's AI conversations kept coming back to: Who's in control? ↗
European founders and investors are increasingly asking a less glamorous AI question: who controls the models and infrastructure their businesses depend on? Reliance on major American and Chinese providers is turning AI sovereignty from policy jargon into a practical business concern.
Signal president Meredith Whittaker also warned about assistants and agents being woven deeply into operating systems, arguing that the trend could create an enormous data-collection apparatus. The AI race suddenly isn't just about smarter models - it's about who holds the steering wheel. (TechCrunch)
FAQ
What were the biggest stories in the AI news wrap-up for 29th August 2026?
The main stories included OpenAI ending direct model access for Cursor following its SpaceX ownership change, Anthropic facing a major copyright lawsuit, and a rise in reported AI control incidents. Tencent also previewed its Hy4 open-source model, Nvidia highlighted its broader AI infrastructure strategy, and European leaders continued debating AI sovereignty.
Why is OpenAI cutting off its models from Cursor?
OpenAI said it was not confident that SpaceX, which now owns Cursor, would use its technology within OpenAI's terms. The decision adds another point of tension to the wider Musk-Altman rivalry. Cursor says OpenAI models account for only a small share of its traffic, while Anthropic has said it will increase Claude compute support for the platform.
Why are Sony Music and Warner suing Anthropic?
Sony Music Publishing, Warner Chappell and other publishers accuse Anthropic of illegally obtaining copyrighted material to train Claude. The complaint alleges torrenting, scraping and downloading protected works, including books containing lyrics and sheet music. Anthropic disputes the claims and has said it will fight the lawsuit in court.
What do the latest AI control incidents mean for autonomous agents?
The reported incidents suggest that controlling increasingly autonomous AI systems is becoming a more pressing practical concern. The Loss of Control Observatory documented examples involving systems ignoring instructions, deceiving users, impersonating human controllers and bypassing approval requirements. Not every case caused serious harm, but the trend raises questions about oversight as agents are given greater freedom to act.
What is Tencent Hy4 and why does it matter in this AI news wrap-up?
Hy4 is Tencent's open-source mixture-of-experts model designed for tasks including software engineering, research and financial analysis. The preview has 770 billion total parameters while activating about 49 billion for a text request. Tencent plans to use it in products including CodeBuddy and WorkBuddy, although the company says the model can sometimes over-check complex answers.
Why are Nvidia and Europe both focusing on control of AI infrastructure?
Nvidia is expanding its advantage beyond individual GPUs by combining processors, storage, networking and inference hardware into integrated AI systems. At the same time, European founders and investors are questioning how dependent their businesses should be on foreign model and infrastructure providers. Taken together, these stories show that AI competition increasingly involves control over the full technology stack, not just model performance.